{
  "text": "就在剛剛,Deepseek API平台正式上線,全新的多模态視覺理解模型,DeepSeek-V4-Flash-Vision-Exp,官方稱其為實驗性質的模型。\n据官方數據,該模型的純文本能力持平DeepSeek-V4-Flash,多模态Agent能力接近OPUS 4.8。\n在API服務中,該模型支持三種主流API格式调用,費用會轉換成Token後按Token計費。\n一張圖片最多占384Token,計費价格与V4Flash模型一致。\n與此同時,官方上線不收費的Files API,用户可先將圖片上傳,再在請求中引用,从而节省带宽。",
  "language": "zh",
  "duration": 38.45225,
  "segments": [
    {
      "id": 0,
      "seek": 0,
      "start": 0.0,
      "end": 10.540000000000001,
      "text": "就在剛剛,Deepseek API平台正式上線,全新的多模态視覺理解模型,DeepSeek-V4-Flash-Vision-Exp,官方稱其為實驗性質的模型。",
      "tokens": [
        50365,
        3111,
        3581,
        32795,
        11,
        11089,
        595,
        405,
        916,
        9362,
        16716,
        15433,
        15789,
        27584,
        5708,
        31699,
        11,
        11319,
        12560,
        1546,
        6392,
        41908,
        3757,
        223,
        27333,
        9463,
        13876,
        17278,
        41908,
        39823,
        11,
        11089,
        595,
        405,
        916,
        371,
        19,
        11,
        25680,
        1299,
        25170,
        33984,
        11,
        31929,
        9249,
        49606,
        9572,
        6344,
        10376,
        24023,
        245,
        21686,
        43450,
        1546,
        41908,
        39823,
        1543,
        50892
      ],
      "temperature": 0,
      "avg_logprob": -0.18096136067011584,
      "compression_ratio": 1.1833333333333333,
      "no_speech_prob": 7.140093637864053e-12,
      "source": "whisper",
      "text_original": "就在剛剛,Deepseek API平台正式上線,全新的多模态視覺理解模型,Deepseek v4,Flash Vision XP,官方稱其為實驗性質的模型。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 1,
      "seek": 0,
      "start": 10.9,
      "end": 18.64,
      "text": "据官方數據,該模型的純文本能力持平DeepSeek-V4-Flash,多模态Agent能力接近OPUS 4.8。",
      "tokens": [
        50910,
        26075,
        106,
        31929,
        9249,
        30622,
        36841,
        11,
        22033,
        41908,
        39823,
        1546,
        7732,
        242,
        17174,
        8802,
        8225,
        13486,
        17694,
        16716,
        11,
        11089,
        595,
        405,
        916,
        371,
        19,
        20232,
        11,
        6392,
        41908,
        3757,
        223,
        32,
        6930,
        8225,
        13486,
        14468,
        17463,
        12059,
        3447,
        1017,
        13,
        23,
        1543,
        51297
      ],
      "temperature": 0,
      "avg_logprob": -0.18096136067011584,
      "compression_ratio": 1.1833333333333333,
      "no_speech_prob": 7.140093637864053e-12,
      "source": "whisper",
      "text_original": "据官方數據,該模型的純文本能力持平,Deepseek v4 Flash,多模态Agent能力接近OPUS 4.8。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 2,
      "seek": 0,
      "start": 18.78,
      "end": 25.26,
      "text": "在API服務中,該模型支持三種主流API格式调用,費用會轉換成Token後按Token計費。",
      "tokens": [
        51304,
        3581,
        4715,
        40,
        27408,
        39517,
        5975,
        11,
        22033,
        41908,
        39823,
        35488,
        10960,
        16516,
        13557,
        27854,
        4715,
        40,
        30921,
        27584,
        8897,
        225,
        9254,
        11,
        1960,
        27364,
        6236,
        32739,
        36338,
        11336,
        51,
        8406,
        5661,
        26613,
        51,
        8406,
        37813,
        50071,
        1543,
        51628
      ],
      "temperature": 0,
      "avg_logprob": -0.18096136067011584,
      "compression_ratio": 1.1833333333333333,
      "no_speech_prob": 7.140093637864053e-12,
      "source": "whisper",
      "text_original": "在API服務中,該模型支持三種主流API格式调用,不便會轉換成Token後按Token擊費。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 3,
      "seek": 2526,
      "start": 25.26,
      "end": 30.660000000000004,
      "text": "一張圖片最多占384Token,計費价格与V4Flash模型一致。",
      "tokens": [
        50365,
        2257,
        18889,
        2523,
        244,
        16668,
        8661,
        6392,
        5322,
        254,
        12625,
        19,
        51,
        8406,
        11,
        37813,
        50071,
        45441,
        30921,
        940,
        236,
        53,
        19,
        25680,
        1299,
        41908,
        39823,
        2257,
        6784,
        112,
        1543,
        50635
      ],
      "temperature": 0,
      "avg_logprob": -0.13851289051335033,
      "compression_ratio": 0.9635416666666666,
      "no_speech_prob": 1.342693491113156e-11,
      "source": "whisper",
      "text_original": "一張圖片最多占384Token,擊費價格与V4Flash模型一致。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 4,
      "seek": 2526,
      "start": 31.040000000000003,
      "end": 38.24,
      "text": "與此同時,官方上線不收費的Files API,用户可先將圖片上傳,再在請求中引用,从而节省带宽。",
      "tokens": [
        50654,
        34127,
        17947,
        13089,
        6611,
        11,
        31929,
        9249,
        5708,
        31699,
        1960,
        18681,
        50071,
        1546,
        37,
        4680,
        9362,
        11,
        9254,
        1486,
        115,
        4429,
        10108,
        34168,
        2523,
        244,
        16668,
        5708,
        40852,
        11,
        8623,
        3581,
        16302,
        32718,
        5975,
        41301,
        9254,
        11,
        35630,
        11070,
        45161,
        2862,
        223,
        4845,
        99,
        2415,
        121,
        1543,
        51014
      ],
      "temperature": 0,
      "avg_logprob": -0.13851289051335033,
      "compression_ratio": 0.9635416666666666,
      "no_speech_prob": 1.342693491113156e-11,
      "source": "whisper"
    }
  ],
  "transcript_selection": {
    "schema": "videonote.transcript_selection.v1",
    "status": "SELECTED",
    "requested_language": "zh",
    "duration": 38.45225,
    "reference_path": null,
    "reference_kind": null,
    "reference_language": null,
    "selected_source": "whisper_segmented",
    "decision_reason": "source_unavailable_or_untrusted_whisper_passed",
    "whisper_rejected": false,
    "source_profile": {
      "segment_count": 0,
      "duration_coverage": 0.0,
      "usable_full": false,
      "usable_partial": false,
      "hallucination_region_count": 0,
      "hallucination_regions": [],
      "language": {
        "requested": "zh",
        "matched": false
      }
    },
    "whisper_profile": {
      "segment_count": 5,
      "multilingual": false,
      "first_start": 0.0,
      "final_end": 38.24,
      "duration_coverage": 0.99448,
      "maximum_gap_seconds": 0.38,
      "inverted_count": 0,
      "language": {
        "requested": "zh",
        "matched": true,
        "latin": 91,
        "han": 134,
        "kana": 0,
        "hangul": 0
      },
      "hallucination_region_count": 0,
      "hallucination_regions": [],
      "usable_full": true,
      "usable_partial": true
    },
    "comparison": {
      "window_seconds": 30.0,
      "compared_window_count": 0,
      "median_score": 0.0,
      "divergent_window_count": 0,
      "divergent_ratio": 0.0,
      "divergent_windows": []
    },
    "timestamp_alignment": {
      "applied": false,
      "estimated_offset_seconds": 0.0,
      "matched_segment_count": 0,
      "match_scores": []
    },
    "caption_timing_repair": null,
    "filled_source_profile": null,
    "whisper_duration_guard": {
      "media_duration": 38.452,
      "rejected_segment_count": 0,
      "rejected_segments": []
    }
  },
  "presegmentation_correction": {
    "applied": true,
    "correction_count": 4,
    "deterministic_fix_count": 0
  }
}